Climate change and the geography of weed damage: Analysis of U.S. maizesystems suggests the potential for significant range transformations

Andrew McDonald a,*, Susan Riha a, Antonio DiTommaso b, Arthur DeGaetano a

a Department of Earth & Atmospheric Sciences, Cornell University, Ithaca, NY 14853, United Statesb Department of Crop & Soil Sciences, Cornell University, Ithaca, NY 14853, United States

Agriculture, Ecosystems and Environment 130 (2009) 131–140

A R T I C L E I N F O

Article history:

Received 24 July 2008

Received in revised form 12 December 2008

Accepted 16 December 2008

Available online 1 February 2009

Keywords:

Biogeography

Global warming

Pests

A B S T R A C T

By the end of the century, climate change projections under a ‘‘business-as-usual’’ emissions scenario

suggest a globally averaged warming of 2.4–6.4 8C. If these forecasts are realized, cropping systems are

likely to experience significant geographic range transformations among damaging endemic weed

species and new vulnerabilities to exotic weed invasions. To anticipate these changes and to devise

management strategies for proactively addressing them, it is necessary to characterize the

environmental conditions that make specific weed species abundant, competitive, and therefore

damaging the production of particular crops (i.e. defining the damage niche). In this study, U.S. maize is

used as a model system to explore the implications of climate change on the distribution of damaging

agricultural weeds. To accomplish this, we couple ensemble climate change projections of annual

temperature and precipitation with survey data of troublesome weed species in maize. At the state scale,

space-for-time substitution techniques are used to suggest the potential magnitude of change among

damaging weed communities. To explore how the geography of damage for specific species may evolve

over the next century, bioclimatic range rules were derived for two weed species that are pervasive in the

Northern (Abutilon theophrasti Medicus, ABUTH) and Southern (Sorghum halepense (L.) Pers., SORHA) U.S.

Results from both analyses suggest that the composition of damaging weed communities may be

fundamentally altered by climate change. In some states, potential changes in the coming decades are

commensurate to those possible by the end of the century. Regions such as the Northeastern U.S. may

prove particularly vulnerable with emerging climate conditions favoring few weed species of present-

day significance. In contrast, regions like the mid-South are likely to experience fewer shifts even with a

similar magnitude in climate change. By the end of the century in the U.S. Corn Belt, cold-tolerant species

like A. theophrasti may be of minor importance whereas S. halepense, a predominantly Southern U.S. weed

species at present, may become common and damaging to maize production with its damage niche

advancing 200–600 km north of its present-day distribution.

� 2008 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment

journal homepage: www.e lsev ier .com/ locate /agee

1. Introduction

1.1. Climate change and agricultural weeds

Under a ‘business-as-usual’ greenhouse gas (GHG) emissionscenario, ensemble climate forecasts project a globally averagedwarming of 2.4–6.4 8C (IPCC, 2007) by the end of the century.Model projections also suggest that temperature increases by mid-century will be only modestly affected by future trends in GHGemissions. With increasing certainty that the Earth’s climate ischanging and that significant warming is inevitable regardless offuture emission reductions, it has become progressively more

* Corresponding author at: 1115 Bradfield Hall, Cornell University, Ithaca, NY

14853, United States. Tel.: +1 607 279 6310; fax: +1 607 255 2106.

E-mail address: ajm9@cornell.edu (A. McDonald).

0167-8809/$ – see front matter � 2008 Elsevier B.V. All rights reserved.

doi:10.1016/j.agee.2008.12.007

important to identify potential vulnerabilities and adaptiveresponses in managed ecosystems (Howden et al., 2007).

Climate change impacts on cropping systems have beenassessed with increasing levels of sophistication for more than30 years (Tubiello et al., 2007). For crop-weed competition, manyexperiments characterize the effects of elevated ambient CO2 oncomparative physiology and growth (e.g. Saebo and Mortensen,1998; O’Donnell and Adkins, 2001; Ziska, 2000, 2001, 2002, 2003),including interactions with factors such as soil nitrogen status (Zhuet al., 2008). Other efforts quantify the role of environmentaldrivers like temperature and water stress on patterns of crop yieldloss from competition (Patterson and Flint, 1979; Patterson et al.,1988; McDonald et al., 2004; Tungate et al., 2007). Indirect impactsof global change may also prove important, with some evidencedemonstrating that herbicide efficacy can be reduced at elevatedCO2 (Harris and Hossell, 2001; Ziska and Teasdale, 2000; Ziskaet al., 2004). Despite the considerable breadth of research

A. McDonald et al. / Agriculture, Ecosystems and Environment 130 (2009) 131–140132

dedicated to understanding potential climate change impacts incropping systems, comparatively little attention has been given topotential effects on the geographic range of agricultural weeds.

Ecological niche theory holds that potential geographicdistribution is governed by the basic environmental requirementsof a species (see Guisan and Thuiller, 2005). This idea is also termedthe conservatism hypothesis: that is, species follow a consistent setof rules in their geographic distribution (Peterson et al., 2003). Thisconcept defines what is referred to as the bioclimatic niche (orenvelope) and establishes the environmental conditions underwhich a species can persist. Environmental factors generallyoperate within a (partially) nested hierarchy with different factorsrelevant at different spatial scales (Pearson and Dawson, 2003).Fossil records and present-day correlative studies demonstratethat climate is the principal determinant of vegetation distributionat regional to global scales (Woodward, 1987, 1988; Patterson,1995). In general, the climate requirements of a species must besatisfied before lower order factors such as topography andlanduse influence spatial distribution (Fig. 1). Potential distribu-tion as delimited by the bioclimatic niche is not equivalent to theactual distribution. Dispersal, disturbance, and competitionprocesses determine which areas encompassed by the bioclimaticniche are actually occupied by a species.

Application of these concepts in cropping systems is not simplytheoretical. In the U.S., Stoller (1973) found that the northern rangelimits of two Cyperaceae weed species corresponded to distinctwinter temperature minima. Across a north–south transect of cerealsystems in Europe, Glemnitz et al. (2000) found that Lapsana

communis L. was found exclusively in the north whereas species suchas Lolium multiflorum Lam. were restricted to the warmer conditionsof Southern Europe. These types of data strongly suggest thatgeographic range transformations for agricultural weeds are highlyprobable outcomes from global climate change (Patterson, 1995;Fuhrer, 2003). If climate change forecasts are realized, croppingsystems are likely to experience a significant change in thegeographic distribution of endemics and, in some regions, anincreased vulnerability to invasion by exotic weed species.

1.2. The ‘damage niche’ concept for agroecosystems

Bioclimatic niche concepts are useful for understanding weeddemography in agroecosystems, but they must be defined morenarrowly when management considerations are the primary

Fig. 1. Hierarchy of resource factors that determine the bioclimatic niche. The bioclimatic

species is influenced by factors such as dispersal, disturbance, and competition proces

objective of a study. Agricultural weed species are typically ofconcern in areas where they are strong competitors rather thansimply persisting at low densities without causing significant cropyield losses. The subjective concept of troublesome integratesenvironment, production, and competition factors to determinegeographic areas where specific weed species tend to be abundantand damaging to crop yield. We introduce the term damage niche torefer to the suite of factors under which specific weed species arejudged troublesome to the production of specific crops.

Fig. 2 illustrates how the damage niche concept in agroeco-systems relates to the bioclimatic niche. Chenopodium album L. is asummer-annual weed that is naturalized across most of NorthAmerica. For the U.S. and Canada, the observed range of C. album isrepresented in grey in Fig. 2. Despite the considerable geographicextent of its bioclimatic niche, this species is only consideredtroublesome to maize in 11 of 38 U.S. states with surveyed maizeproduction systems (black circles, Fig. 2). From the clusteredspatial distribution of these states, it is apparent that precipitationand temperature are both likely candidates for defining theboundaries of the damage niche for this species in maize. Ingeneral, C. album is not judged troublesome to maize under thewarmer conditions of the Southern U.S. or the drier conditions ofthe western U.S.

1.3. Projecting weed distributions in a changing climate

The most widely used analytical approaches for predictingfuture species distributions with climate change are bioclimaticniche models (BNM). These biogeographic tools apply statistical ormachine-learning methods for quantifying associations betweensurveyed species distributions and environmental factors. Exam-ples include CLIMEX (Sutherst and Maywald, 1985), GARP(Stockwell and Peters, 1999), SPECIES (Pearson et al., 2002),BIOMAPPER (Hirzel et al., 2002), and BIOMOD (Thuiller, 2003).BNM may provide a robust methodology for quantifying thedamage niche for agricultural weeds (see Section 1.2). At present,however, surveys of troublesome weeds in cropping systems arelimited with respect to geographic coverage and spatial resolution.For the U.S., Bridges (1992) canvassed expert judgment to compilelists of troublesome weed species for major crops in each state. Torun a BNM model like GARP, a minimum of 15–20 speciesoccurrence points are required, and this standard does not includedata for model validation (Raimundo et al., 2007). With states

niche establishes the potential geographic range for a species. The realized range of a

ses.

Fig. 2. The geographic range for C. album includes almost all regions of the U.S. and Canada (grey shaded areas of the map). Within the U.S., states with maize production that

were surveyed for troublesome weed species by Bridges (1992) are indicated with circles. Despite its extensive geographic range, C. album was only judged troublesome to

maize production in states with black circles. This map illustrates that the damage niche for C. album in maize is much narrower than its bioclimatic niche which governs

overall geographic range. (Distribution map for C. album adopted from USDA’s PLANTS database, http://plants.usda.gov/.)

A. McDonald et al. / Agriculture, Ecosystems and Environment 130 (2009) 131–140 133

treated as ‘points’, this minimum requirement is met for only oneof the more than 60 species identified by Bridges (1992) astroublesome to maize production.

In the absence of higher-resolution weed survey data thanprovided by Bridges (1992), this project uses space-for-time (SFT)substitution and maize as a model system to explore howtroublesome weed communities in different U.S. states may evolvewith projected changes in mean annual temperature and precipita-tion.For global change research, SFT identifies present-day analoguesfor projected climate conditions in order to characterize potentialecosystem responses (Ziska, 2003; Carreiro and Tripler, 2005). Inother words, SFT infers the impacts of climate change from currentbiogeographical patterns in the landscape. For this project, estimatesof mean annual precipitation and temperature were projected fortwo 30-year periods: 2030 (i.e. 2016–2045, ‘coming decades’) and2084 (i.e. 2070–2099, ‘end of century’). To assess how the geographyof damage for individual weed species may be altered by climatechange, we also derive simple climate-based range rules that definethe damage niche for the two species (Abutilon theophrasti, Sorghum

halepens) that are the most prevalent troublesome species inNorthern and Southern U.S. maize systems, respectively.

2. Materials and methods

2.1. Weed survey data

In the early 1990s, Bridges (1992) canvassed expert judgmentto compile lists of the 10 most troublesome weed species in major

cropping systems for each U.S. state. The concept of troublesomeintegrates both weed abundance and capacity to cause substantialcrop yield losses. With this methodology, weed species are alsoassigned a numerical rank from most (1) to least troublesome (10).In the Bridges (1992) survey, troublesome weed communities inmaize were assessed in 38 U.S. states. The number of weed speciescharacterized as troublesome was capped at 10, but fewer than 10species were reported for some states. Bayer codes (now referredto as EPPO codes, see http://eppt.eppo.org) are used by Bridges(1992) to identify species. For several genera (Amaranthus,Cenchrus, Cyperus, Digitaria, Ipomoea, Rubus, Setaria, and Solanum),species were not differentiated in all states. To facilitate cross-statecomparisons in this study, species-level distinctions were notconsidered for these genera.

2.2. Climate data

Hayhoe et al. (2008) have developed statistically downscaledU.S. climate projections at a spatial resolution 1/88 (ca. 140 km2).These projections are based on several atmosphere-ocean globalcirculation model (AOGCM) forecasts under the IPCC’s SRES high(A1fi), mid-high (A2) and low (B1) greenhouse gas emissionscenarios (Nakicenovic et al., 2000). Different greenhouse gasemission scenarios reflect diverse development pathways withrespect to several socio-economic factors including populationgrowth and technological change. From a current atmosphericconcentration of approximately 385 ppm, CO2 concentrations areprojected to reach 550 and 970 ppm under the low (B1) and high

A. McDonald et al. / Agriculture, Ecosystems and Environment 130 (2009) 131–140134

(A1fi) emission scenarios, respectively, by the end of the century.Global temperature projections for different emission scenariosare similar until approximately 2050 (IPCC, 2007), suggesting thatsome changes in climate are inevitable regardless of efforts toreduce GHG emissions.

For this study, we use the A1fi scenario, commonly referred toas ‘business-as-usual’ GHG emissions, to forecast climate changesuntil the end of the century. Monthly temperature and precipita-tion projections were derived from three different AOGCMs: GFDLCM2.1 (Delworth et al., 2006), HadCM3 (Pope et al., 2000), andPCM1 (Washington et al., 2000). An ensemble forecast of meanannual temperature and precipitation was then computed bysimple averaging of AOGCM output. From the ensemble forecast,future climatology (i.e. 30-year weather averages) centered on2030 (i.e. 2016–2045, ‘coming decades’) and 2084 (i.e. 2070–2099,‘end of century’) was predicted. Historical climatology (1961–1990) for annual precipitation and temperature was based onmonthly observations from the United States Historical Climatol-ogy Network gridded to the same 1/88 spatial resolution as theAOGCM projections (see Hayhoe et al., 2008). For future andhistorical climatology, area-wide mean values for annual pre-cipitation and temperature were calculated for each U.S. state.Based on these calculations, we identified close historicalanalogues for projections of future climatology (i.e. precipitationdifference �10 cm with temperature difference �0.6 8C).

2.3. Weed community comparisons

Within the space-for-time substitution and climate analogueapproach, the Bray-Curtis (BC) dissimilarity metric for multivariatedata was used to make pair-wise comparisons of troublesomeweed communities between U.S. states. For this purpose, specieswere inversely weighted by their Bridges (1992) ranking from themost troublesome species (10) to the least (1). Species that werenot judged troublesome in a state were assigned a value of zero. BCis well suited for multivariate comparisons when the objects (e.g.U.S. states) have many zero values among the variables (e.g.species) (Mac Nally, 1989; Quinn and Keough, 2002). The BC metricvaries from 0 to 1, with 1 indicating 100% dissimilarity betweenobjects.

To simultaneously contrast weed community compositionacross all 38 states that have weed survey data for maize, a BCdissimilarity matrix for the full dataset was subjected to principalcoordinate analysis (PCoA). PCoA translates multivariate dissim-ilarities between objects into Euclidean distances (Quinn andKeough, 2002). To aggregate the state-based weed communitiesinto mega-groups, agglomerative hierarchical cluster analysis wasperformed with the first two coordinates of the PCoA. The finalpartition was constrained by the pre-analysis specification of fourclusters, a number suggested by visual evaluation of PCoA output.Linear discriminant analysis with cross-validation was used toquantify the association between state-based climate parametersand membership in the different groups.

The Bray-Curtis dissimilarity and PCoA analyses were con-ducted with the PAST software package (http://folk.uio.no/ohammer/past/), and Minitab 15 was used for cluster anddiscriminant analyses.

2.4. Developing bioclimatic envelopes (i.e. range rules) for the damage

niche

Climate-based range rules for the damage niche were derivedfor S. halepense and A. theophrasti by identifying maximum andminimum values of annual temperature and precipitation amongU.S. states where these species are characterized as troublesome tomaize production by Bridges (1992). Since crop water availability

could not be quantified for states dominated by irrigatedproduction practices (i.e. operationally defined as >50% of maizeacreage as reported in NASS, 1992), these states were excludedfrom our analysis. For S. halepense, climate data were used from thefollowing 15 states: AL, AR, FL, GA, IL, IN, KY, LA, MD, MS, MO, NC,SC, TN, and WV. For A. theophrasti, climate data were used from thefollowing 10 states: IN, IA, MI, MO, NJ, NY, OH, PA, WV, and WI.Neither species is damaging to maize production in the drierwestern states in the absence of irrigation. Damage niche rangeboundaries for temperature act in opposite directions, with S.

halepense limited by the cooler conditions in the Northern U.S. andA. theophrasti limited by the warmer conditions in the south. Twodifferent criteria were used to establish range rules from state-based historical climatology. In the limiting direction (e.g. coolertemperatures for S. halepense), the mean climate value for the stateat the extreme was used. For example, the state with the coolestmean annual temperature where S. halepense is damaging to maizeproduction is West Virginia (WV) and the lower limit of the rangerule was set at the mean annual temperature for WV. In the otherdirection, where there is no apparent climate limitation in theconterminous U.S. (e.g. warm states for S. halepense), the range rulewas set at the maximum (or minimum) climate value in the state atthe extreme (e.g. warmest region in the warmest state for S.

halepense). Following the climate envelope method developed byNix (1986), all values that fall between the maximum andminimum are encompassed by the range rules. Due to theaforementioned limitations of state-scale weed survey data, thepredictive accuracy of these rules could not be assessed.

Range rules were projected on a map of historical (1961–1990)U.S. climatology at the scale of the climate grids (i.e. 140 km2) toestablish the contemporary geographic extend of the damageniche using ArcGIS1 9 geoprocessing software. To assess how thegeography of the damage niche may evolve with forecastedclimate changes under the ‘business-as-usual’ emissions scenario,range rules were subsequently projected onto maps of future U.S.climatology centered on 2030 (i.e. 2016–2045, ‘coming decades’)and 2084 (i.e. 2070–2099, ‘end of century’). Since no effort is madeto modify these rules with soil, terrain, or other non-climatecriteria, they should be interpreted as coarse-scale (i.e. regional)indicators of potential geographic distribution.

3. Results

3.1. Assessing the potential for weed community change with climate

analogues and SFT substitution

We identified nine U.S. states (AL, DE, IN, KY, MI, NJ, NY, PA, andSC) with present-day analogues to projected climate changes thatalso have weed survey data for maize (Table 1). Table 2 lists weedspecies that are currently considered damaging and contrasts themwith species judged damaging in states that are analogues toclimate projections centered on 2030 (2016–2045) and 2084(2070–2099) under ‘business-as-usual’ GHG emissions. Speciesthat are likely to remain damaging to maize are highlighted in boldwith the total number of original species retained in eachtimeframe reported at the bottom of the columns.

Our results suggest that the types of weed community changesin maize are unlikely to be similar across all states. For example, inNew York (NY) none of the species currently considered damagingto maize are damaging in Kentucky (KY), the state that New York’sannual climate is projected to resemble in 2084. Conversely, inSouth Carolina (SC) 7 of 10 species now considered damaging tomaize are also damaging in Florida (FL), the state that SC isexpected to resemble towards the end of the century. Elsewhere,expected changes fall between these extremes. For Kentucky, S.

halepense is considered the most damaging weed species in maize

Table 1Recent and projected climatology for selected U.S. states under a ‘business-as-usual’ emissions scenario. These states have survey data for troublesome weed species in maize

and also historical analogues for projected changes to climate (right side of table). A list of state abbreviations is published by the U.S. Postal Service (http://www.usps.com/

ncsc/lookups/usps_abbreviations.html).

Historical and projected annual climatology (A1fi – ‘business-as-usual’ emissions scenario) Historical analogues (for projected climatology)

State Period Temperature (8C) Precipitation (cm) State Temperature (8C) Precipitation (cm)

AL 1961–1990 16.9 138 — — —

2016–2045 18.5 147 LA 18.9 141

2070–2099 22 142 FL 21.5 136

DE 1961–1990 13.1 110 — — —

2016–2045 14.6 119 NC 14.9 122

2070–2099 17.9 125 GA 17.5 125

IN 1961–1990 10.8 100 — — —

2016–2045 12.7 107 VA 12.9 108

2070–2099 16.5 115 SC 17 118

KY 1961–1990 12.9 119 — — —

2016–2045 14.6 128 NC 14.9 122

2070–2099 18.3 137 LA 18.3 137

MI 1961–1990 6.9 81 — — —

2016–2045 8.7 83 IA 8.9 83

2070–2099 12.4 90 MO 12.5 100

NJ 1961–1990 11.3 115 — — —

2016–2045 12.9 126 KY 12.9 119

2070–2099 16.4 132 AL 16.9 138

NY 1961–1990 7.1 102 — — —

2016–2045 8.8 110 PA 8.9 106

2070–2099 12.5 115 KY 12.9 119

PA 1961–1990 8.9 106 — — —

2016–2045 10.7 113 WV 10.5 111

2070–2099 14.3 120 NC 14.9 122

SC 1961–1990 17 118 — — —

2016–2045 18.5 131 LA 18.9 141

2070–2099 21.7 135 FL 21.5 136

A. McDonald et al. / Agriculture, Ecosystems and Environment 130 (2009) 131–140 135

at present and is also the most damaging species in the states thatKentucky may resemble in 2030 (NC) and 2084 (LA). However,none of the other species currently considered damaging inKentucky are likely to remain so by the end of the century.

A measure of multivariate dissimilarity (Bray-Curtis) betweenthe current weed community in a state and that of its projectedclimate analogues for 2030 and 2084 is presented in Fig. 3. Withthis measure, a value of 1 indicates 100% dissimilarity betweenweed communities. Three results are noteworthy when consider-ing trends among the states. First, potential changes in comingdecades for several states (i.e. AL, SC, MI, KY, and DE) are similar to

Fig. 3. For maize systems in selected U.S. states, Bray-Curtis dissimilarity between

current communities of troublesome weed species and communities that are

projected to be favored by climate conditions in 2030 and 2084. A value of 1

indicates complete dissimilarity between weed communities.

those possible by the end of the century. Second, there are strongregional differences with states in the Northeastern U.S. (i.e. NY, NJ,DE, and PA) predicted to experience more extensive changes intheir damaging weed communities than states in the Southern U.S.(i.e. Al, SC). Differences between geographic regions are not relatedto a greater degree of climate change (see Table 1). Rather, states inthe Northeastern U.S. are projected to cross a climate transitionzone that separates weed communities with substantially differentcompositions whereas states like Alabama (AL) and South Carolina(SC) are not expected to cross any major transition zones (seeSection 3.2). Lastly, in some regions large weed communitychanges are likely by the end of the century with states like NY, NJ,PA, and DE projected to have climate conditions that will favor anentirely different suite of troublesome weed species than atpresent (i.e. BC > 0.85).

To evaluate the reliability of this SFT approach for predictingfuture weed community composition based on state-scale climatefactors, we identified six pairs of states that are contemporaryclimate analogous and compared their troublesome weed com-munities by computing the Bray-Curtis dissimilarity metric foreach pair. Excluding one case which was an outlier (KY and DE), themean BC value for these comparisons was 0.53 (SE � 0.05) and thepairs share, on average, 5 species in common with a range from 4 to 6.This indicates that our approach provides a reasonable, but notperfect, methodology for predicting future weed communitycomposition when applied at the scale of U.S. states.

3.2. Weed community mega-groups and associations with climate

In order to explore how troublesome weed communities varyacross all 38 U.S. states surveyed by Bridges (1992), BC

Table 2For U.S. states, weed species (EPPO codes) that are currently considered damaging

to maize production contrasted to those judged damaging in states that are close

analogues to projected changes to climate centered on 2030 (2016–2045) and 2084

(2070–2099). Species that are expected to remain climatically favored are

highlighted in bold. Full scientific names for EPPO codes can be accessed at

http://eppt.eppo.org/.

Alabama (AL) Present 2030 2084

(AL historical) (LA historical) (FL historical)

Weed rank

1 SORHA SORHA PANTE2 PANTE ROOEX DEDTO

3 BRAPP BRAPP CASOB4 IPO sp. IPO sp. SORHA5 PANDI XANST

6 CASOB IPO sp.

7 AMA sp. ACNHI

8 CYP sp. CYP sp.

9 AMA sp.

10 DIG sp.

RETAINED: 3/8 6/8

Delaware (DE) Present 2030 2084

(DE historical) (NC historical) (GA historical)

Weed rank

1 CIRAR SORHA PANTE

2 PANDI PANTE IPO sp.

3 APCCA BRAPP XANST

4 SET sp. CASOB CASOB

5 AMA sp. IPO sp. CASOC

6 CYP sp. SORHA

7 SIYAN

8 SOL sp.

9 ABUTH

10 CYNDA

RETAINED: 0/5 0/5

Kentucky (KY) Present 2030 2084

(KY historical) (NC historical) (LA historical)

Weed rank

1 SORHA SORHA SORHA2 SORVU PANTE ROOEX

3 AMBTR BRAPP BRAPP

4 AMPAL CASOB IPO sp.

5 PANDI IPO sp.

6 SIYAN CYP sp.

7 CMIRA SIYAN8 CONAR SOL sp.

9 IPO sp. ABUTH

10 XANST CYNDA

RETAINED: 3/10 2/10

Indiana (IN) Present 2030 2084

(IN historical) (VA historical) (SC historical)

Weed rank

1 ABUTH No data CYNDA

2 AMBTR PANTE

3 SORHA BRAPP

4 CIRAR SORHA5 XANST IPO sp.

6 SET sp. CYP sp.

7 IPO sp. CASOB

8 SIYAN PANDI

9 APCCA AMA sp.

10 DATST XANST

RETAINED: — 3/10

Michigan (MI) Present 2030 2084

(MI historical) (IA historical) (MO historical)

Weed rank

1 ABUTH SET sp. ABUTH2 PANDI ABUTH SORVU

Table 2 (Continued )

Michigan (MI) Present 2030 2084

(MI historical) (IA historical) (MO historical)

3 AGRRE AMA sp. SET sp.

4 CHEAL CHEAL AMA sp.

5 CIRAR XANST PANDI6 APCCA POLPY ASCSY

7 CONAR HELAN APCCA8 DIG sp. SORVU SORHA

9 SET sp. ERBVI CHEAL10 AGRRE XANST

RETAINED: 4/9 5/9

New York (NY) Present 2030 2084

(NY historical) (PA historical) (KY historical)

Weed rank

1 ABUTH AMBEL SORHA

2 CHEAL APCCA SORVU

3 MUHFR SOL sp. AMBTR

4 ASCSY MUHFR AMPAL

5 SOL sp. RUB sp. PANDI

6 CAGSE AGRRE SIYAN

7 SET sp. ABUTH CMIRA

8 SIYAN CONAR

9 CONAR IPO sp.

10 CHEAL XANST

RETAINED: 4/7 0/7

New Jersey (NJ) Present 2030 2084

(NJ historical) (KY historical) (AL historical)

Weed rank

1 APCCA SORHA SORHA

2 SORVU SORVU PANTE

3 ABUTH AMBTR BRAPP

4 AMPAL IPO sp.

5 PANDI PANDI

6 SIYAN CASOB

7 CMIRA AMA sp.

8 CONAR CYP sp.

9 IPO sp.

10 XANST

RETAINED: 1/3 0/3

Pennsylvania (PA) Present 2030 2084

(PA historical) (WV historical) (NC historical)

Weed rank

1 AMBEL SORHA SORHA

2 APCCA AMA sp. PANTE

3 SOL sp. CHEAL BRAPP

4 MUHFR ABUTH CASOB

5 RUB sp. MUHFR IPO sp.

6 AGRRE AGRRE CYP sp.

7 ABUTH CYP sp. SIYAN8 SIYAN ASCSY SOL sp.

9 CONAR APCCA ABUTH10 CHEAL SIYAN CYNDA

RETAINED: 6/10 3/10

S. Carolina (SC) Present 2030 2084

(SC historical) (LA historical) (FL historical)

Weed rank

1 CYNDA SORHA PANTE2 PANTE ROOEX DEDTO

3 BRAPP BRAPP CASOB4 SORHA IPO sp. SORHA5 IPO sp. XANST6 CYP sp. IPO sp.

7 CASOB ACNHI

8 PANDI CYP sp.

9 AMA sp. AMA sp.

10 XANST DIG sp.

RETAINED: 3/10 7/10

A. McDonald et al. / Agriculture, Ecosystems and Environment 130 (2009) 131–140136

Fig. 4. Principal coordinate analysis (PCoA) of troublesome weed communities in

maize for all 38 U.S. states surveyed by Bridges (1992). Euclidean distances indicate

the degree of similarity between weed communities. Cluster analysis was used to

divide the weed communities into four mega-groups based on the first two PCoA

coordinates. The general geographic region encompassed by these groups is noted

in the figure key.

Fig. 5. Box plots of mean annual temperature (8C) for U.S. states that belong to each

of the four weed community groups identified in Fig. 4. The middle line in the boxes

is the median value with the 25% quartiles indicated by ends of the box and the most

extreme values by the whiskers.

A. McDonald et al. / Agriculture, Ecosystems and Environment 130 (2009) 131–140 137

dissimilarities between states were translated into Euclideandistances with PCoA. The first two coordinates, explaining 35% ofthe total variance between states, are presented in Fig. 4. Fourdistinct mega-groups emerge which can be roughly generalizedgeographically as: (1) Northern Corn Belt/Western U.S., 2) Mid-Atlantic/Central Corn Belt, (3) Southern Corn Belt, and (4) SouthernU.S. As determined by average damage rank, the following weedspecies most strongly characterize each group and are listed indescending order of importance: Group 1: Panicum miliaceum L.,Cirsium arvense (L.) Scop., Setaria species, and Elytrigia repens (L.)Nevski); Group 2: A. theophrasti, Setaria species, Apocynum

cannabinum L., and Sorghum bicolor (L.) Moench; Group 3: S.

halepense, S. bicolor, Ambrosia trifida L., and A. theophrasti; Group 4:S. halepense, Panicum texanum Buckl., Ipomoea species, andUrochloa platyphylla (Nash) R.D. Webster. Note that Group 3shares troublesome weed species with Groups 2 and 4 andtherefore represents a region of transition and overlap rather thanan entirely unique suite of damaging weed species.

A key point derived from Fig. 4 is that climate differences arenon-linear predictors of weed community differences betweenstates. For example, the annual temperature climatology ofTennessee (13.9 8C, TN) is approximately equidistant betweenOhio (10.2 8C, OH) and Georgia (17.5 8C, GA). In contrast, from theEuclidean distances separating troublesome weed communities inFig. 4, it is clear that communities in TN are very similar to those inGA and very different than those in OH. Hence comparable changein climate may have quite diverse, location-dependent impacts onweed community composition.

For predicting the potential impact of climate change ontroublesome weed communities in general terms, it is useful toidentify climate thresholds that segregate major community types.Fig. 5 presents box and whisker plots for mean annual temperatureamong states that belong to the four major groups identified inFig. 4. Despite the very coarse spatial and temporal resolution ofstate-based mean annual temperature, there is very little overlapbetween Groups 1, 2, and 4 in Fig. 5. For example, all 12 states with

Table 3Damage niche range rules for A. theophrasti and S. halepense. These rules define the range

where these weed species have historically been judged troublesome to maize producti

climate values that set the range limits for each species.

Maximum temperature (8C) Minimum temperature (

A. theophrasti 12.4 (MO mean) 3.7 (WI minimum)

S. halepense 24.5 (FL maximum) 10.5 (WV mean)

mean annual temperatures above 13.2 8C belong to weed Group 4(i.e. ‘Southern U.S.’). Groups 1, 2, and 4 represent very distinct weedcommunity types, whereas Group 3 shares attributes of Groups 1and 2 and is less readily distinguished on the basis of annualtemperature. Overall, linear discriminant analysis demonstratesthat state-based mean annual temperature can be used to correctlypredict weed community types for 29 out of 38 states (76%accuracy). The temperature thresholds derived from this analysissuggest proximate weed community transitions at 8.1 8C (Group1! 2), 10.8 8C (Group 2! 3), and 14.1 8C (Group 3! 4).

3.3. Damage niche range transformations for individual species

Damage niche range rules for S. halepense and A. theophrasti

based on annual precipitation and temperature climatology arereported in Table 3. These species are troublesome to maizeproduction primarily in the eastern half of the U.S., with S.

halepense damage to maize restricted to regions where meanannual temperatures exceed�10.5 8C and A. theophrasti restrictedto regions below �12.4 8C. Based on the range rules presented forthese species in Table 3, the historical distribution of the damageniche in the conterminous U.S. and projected future distributionsunder a ‘business-as-usual’ climate change scenario for 30-yearclimatology centered on 2030 and 2084 are presented in Figs. 6and 7.

At present, A. theophrasti is damaging to maize productionacross the Great Lake States, Corn Belt, Mid-Atlantic, and North-eastern States. For 2030, our projections suggest a 100–300 kmpole-ward migration of conditions that favor A. theophrasti damageto maize. Near the end of this century, this pole-ward retreat mayextend approximately 200–650 km north of present-day bound-aries and A. theophrasti may only be damaging to maize in thenorthern portions of the Northeast (i.e. Vermont, New York) and inthe Great Lake States (i.e. Minnesota, Wisconsin, Michigan).

In contrast to A. theophrasti, S. halepense will likely expand itshistorical range of damage to U.S. maize with projected changes to

of annual climate conditions (30-year averages for precipitation and temperature)

on. In parentheses are the U.S. state abbreviations for the geographic sources of the

8C) Maximum precipitation (cm) Minimum precipitation (cm)

129.4 (NJ maximum) 80.3 (MI mean)

161.7 (LA maximum) 94.1 (IL mean)

Fig. 6. Historical and projected distribution of the damage niche for A. theophrasti in U.S. maize cropping systems. Projections are for climatology centered on 2030 and 2084

under a ‘business-as-usual’ GHG emission scenario. Towards the end of the century, the damage niche for A. theophrasti may experience a pole-ward retreat of approximately

200–650 km north of present-day boundaries.

A. McDonald et al. / Agriculture, Ecosystems and Environment 130 (2009) 131–140138

climate. At present, S. halepense is not judged troublesome in theNorthern Corn Belt, Northeastern States, or Great Lakes States. Incoming decades, the damage niche will likely extend throughmuch of the Corn Belt and into southern portions of the

Fig. 7. Historical and projected distribution of the damage niche for S. halepense in U.S. m

under a ‘business-as-usual’ GHG emission scenario. Towards the end of the century, the d

200–600 km north of present-day boundaries.

Northeastern States. By the end of the century, the damage nichewill encompass much of the Northeast and southern parts of theLake States. Overall, the pole-ward advance of the S. halepense

damage niche will likely extend from 200 to 600 km beyond its

aize cropping systems. Projections are for climatology centered on 2030 and 2084

amage niche for S. halepense may experience a pole-ward advance of approximately

A. McDonald et al. / Agriculture, Ecosystems and Environment 130 (2009) 131–140 139

historical boundaries towards the end of the century under the‘business-as-usual’ GHG emission scenario.

Species distributions do not conform to political boundaries andit is clear that in several states where a species has been judgedtroublesome that this designation does not hold for every locationwithin that state. The opposite is also true, with some species notconsidered troublesome at the state-scale that are damaging tocrop production in smaller regions within a state. For mapping thehistorical extent of the damage niche, our method results inpredictions that alternately appear to both over and under-predictthe geographic range of the damage niche. In the case of S.

halepense, our rule encompasses 86% of the land area in stateswhere this species was judged troublesome; on the other hand,approximately 21% of the overall extent is in states where S.

halepense is not currently judged damaging to maize. Theseprojections may prove to be an accurate depiction of reality, butthis cannot be assessed without weed survey data at a finer scalethan what is provided by Bridges (1992). Hence, our projections ofhistorical and future damage ranges are best viewed in acomparative sense and not as precise predictions.

4. Discussion

It is important to emphasize that the results of this studysuggest how the geographic distribution of troublesome weeds inU.S. maize will potentially evolve in a changing climate. Manyfactors other than climate substantially influence actual speciesdistributions including competitive exclusion (Mack, 1996; Daviset al., 1998a,b), dispersal limitations (Lawton, 2000), and patternsof disturbance (Guisan and Thuiller, 2005). That acknowledged,annual cropping systems have several attributes that may makeclimate considerations particularly important for predicting futureweed distributions. Activities such as tillage and crop harvest arerelatively uniform and predictable perturbations. Ecosystems withhigh levels of disturbance are more vulnerable to colonization bynewly introduced plant species and are likely to reach acomparatively rapid equilibrium with emergent climate factors(Hobbs and Huenneke, 1992; Milchunas and Lauenroth, 1995).Further, weed dispersal processes are facilitated by the high levelof habitat continuity in major cropping systems like maizeproduction in the U.S. and also by vectors like tillage, manurespreading, and seed exchanges that facilitate seed movementwithin and between farms (Cousens and Mortimer, 1995).Moreover, many weed species that are climatically favored tobecome troublesome in new regions are already present in thelandscape, even though they are not damaging at present (seeFig. 1). For these species, dispersal processes will not limit damageniche range transformations in a changing climate.

There are, however, other challenges to predict the potentialimpact of climate changes on agricultural weeds that must beacknowledged. Agronomic practices for particular crops are notstatic in time and space; new classes of herbicides, cultivars, tillageinnovations, use of irrigation, and seed cleaning practices can allinfluence the geographic distribution and crop damage caused byagricultural weeds (Salisbury, 1961; Froud-Williams et al., 1984;Clements et al., 1996). For example, evidence suggests that therecent introduction of glyphosate resistant crops can significantlychange weed community composition (Harker et al., 2005). Sincethe development and use of different agricultural practices ishighly unpredictable, there is an inherent element of uncertaintyto the use of bioclimatic envelopes or space-for-time substitutionfor projecting future weed distributions. Also, the possibility thatagricultural weed populations will evolve new traits in response toemerging climate and non-climate selection pressures cannot bediscounted (Clements et al., 2004). Perhaps most importantly,biogeographic methods do not account for the impact of evolving

atmospheric chemistry on competitive interactions. Any environ-mental change that differentially affects the morphology, growth,or reproduction of interacting plant communities has the potentialto modify the spatial extent of the damage niche (Patterson, 1995;Bunce, 2001). However, since maize possesses the C4 photosyn-thetic pathway and does not respond dramatically to CO2

enrichment, it is likely that most weed species will either becomemore competitive (C3) or maintain similar competitive abilities(C4) at elevated CO2 concentrations (Patterson, 1995; Pattersonet al., 1999). Hence, if a weed is presently characterized asdamaging to maize under a certain set of environmentalconditions, it is likely that it will remain so as atmospheric CO2

increases.Potential changes in the weed biogeography of agricultural

systems pose a challenge to management, but also an opportunity.If weed species can be identified as favored due to emergentclimate conditions in a given region, nascent populations can betargeted for control before they become well established. Thisstudy can be viewed as a ‘proof of concept’ and a first step towardsdeveloping this type of information for major cropping systems.Finer-scale survey data for the present-day geography of weeddamage would enable more quantitative and spatially resolvedpredictions of potential range transformations in a changingclimate.

5. Conclusion

With U.S. maize as a model system, our results suggest that thecommunity composition of damaging agronomic weeds may befundamentally transformed by climate change. In some U.S. states,potential changes in coming decades are similar to those possibleby the end of the century. Regions such as the Northeastern U.S.may prove particularly vulnerable, with future climates projectedto favor few weed species of present-day significance. Otherregions are likely to experience rather minor weed communityshifts, even with a similar magnitude of climate change. For a givenregion, potential community impacts appear to be chieflycontingent on proximity to climate transition zones that separatemajor weed community types. For individual species, pole-wardmigration of the damage niche may be on the order of 200–600 kmby the end of the century. If weed species can be identified asfavored due to emergent climate conditions in a given region,expanding or newly introduced populations can be targeted forcontrol before they become well established. The accuracy of thesetypes of projections can be refined by collecting finer-scale surveydata for troublesome weeds species in major cropping systems.

Acknowledgements

The authors wish to thank Katharine Hayhoe for provingdownscaled climate change data and Brian Belcher for assistancewith data processing. The corresponding author would also like tothank the Northeastern Weed Science Society for organizing asymposium on climate change and weeds which was instrumentalto the completion of this work.

References

Bridges, D.C. (Ed.), 1992. Crop Losses due to Weeds in the United States—1992.Weed Science Society of America, Champaign, IL.

Bunce, J.A., 2001. Weeds in a changing climate. In: Riches, C.R. (Ed.), Weed Manage-ment Constraints under Climate Change. British Crop Protection Council, Farn-ham, UK.

Carreiro, M.M., Tripler, C.E., 2005. Forest remnants along urban-rural gradients:examining their potential for global change research. Ecosystems 8, 568–582.

Clements, D.R., Benoit, D.L., Murphy, S.D., Swanton, C.J., 1996. Tillage effects onweed seed return and seedbank composition. Weed Science 44, 314–322.

A. McDonald et al. / Agriculture, Ecosystems and Environment 130 (2009) 131–140140

Clements, D.R., DiTommaso, A., Jordan, N., Booth, B.D., Cardina, J., Doohan, D., Mohler,C.L., Murphy, S.D., Swanton, C., 2004. Adaptability of plants invading NorthAmerican cropland. Agriculture, Ecosystems, and Environment 104, 379–398.

Cousens, R., Mortimer, M., 1995. Dynamics of Weed Populations. Cambridge Uni-versity Press, Cambridge.

Davis, A.J., Jenkinson, L.S., Lawton, J.L., Shorrocks, B., Wood, S., 1998a. Makingmistakes when predicting shifts in species range in response to global warming.Nature 391, 783–786.

Davis, A.J., Lawton, J.L., Shorrocks, Jenkinson, L.S., 1998b. Individualistic speciesresponses invalidate simple physiological models of community dynamicsunder global environmental change. Journal of Animal Ecology 67, 600–612.

Delworth, T.L., Broccoli, A.J., Rosati, A., Stouffer, R.J., Balaji, V., Beesley, J.A., Cooke,W.F., Dixon, K.W., Dunne, J., Dunne, K.A., Durachta, J.W., Findell, K.L., Ginoux, P.,Gnanadesikan, A., Gordon, C.T., Griffies, S.M., Gudgel, R., Harrison, M.J., Held,I.M., Hemler, R.S., Horowitz, L.W., Klein, S.A., Knutson, T.R., Kushner, P.J.,Langenhorst, A.R., Lee, H.C., Lin, S.J., Lu, J., Malyshev, S.L., Milly, P.C.D., Ramas-wamy, V., Russell, J., Schwarzkopf, M.D., Shevliakova, E., Sirutis, J.J., Spelman,M.J., Stern, W.F., Winton, M., Wittenberg, A.T., Wyman, B., Zeng, F., Zhang, R.,2006. GFDL’s CM2 global coupled climate models. Part 1. Formulation andsimulation characteristics. Journal of Climate 19, 643–674.

Froud-Williams, R.J., Chancellor, R.J., Drennan, D.S.H., 1984. The effects of seedburial and soil disturbance on emergence and survival of arable weeds inrelation to minimal cultivation. Journal of Applied Ecology 21, 629–641.

Fuhrer, J., 2003. Agroecosystem responses to combinations of elevated CO2, ozone,and global climate change. Agriculture Ecosystems & Environment 97, 1–20.

Glemnitz, M., Czimber, G., Radics, L., Hoffmann, J., 2000. Weed flora compositionalong a north-south climate gradient in Europe. Acta Agronomica Ovariensis 42,155–169.

Guisan, A., Thuiller, W., 2005. Predicting species distribution: offering more thansimple habitat models. Ecology Letters 8, 993–1009.

Hayhoe, K., Wake, C.P., Anderson, B., Liang, X.Z., Maurer, E., Zhu, J., Bradbury, J.,DeGaetano, A., Stoner, A.M., Wubbles, D., 2008. Regional climate change projec-tions for the Northeast USA. Mitigation and Adaptation Strategies for GlobalChange 13, 425–436.

Harker, K.N., Clayton, G.W., Blackshaw, R.E., O’Donovan, J.T., Newton, Z.L., Johnson,E.N., Gan, Y., Zentner, R.P., Lafond, G.P., Irvine, R.B., 2005. Glyphosate-resistantspring wheat production system effects on weed communities. Weed Science53, 451–464.

Harris, D., Hossell, J.E., 2001. Weed management constraints under climate change.In: Riches, C.R. (Ed.), Weed Management Constraints under Climate Change.British Crop Protection Council, Farnham, UK.

Hirzel, A.H., Hausser, J., Chessel, D., Perrin, N., 2002. Ecological-niche factor analysis:how to compute habitat-suitability maps without absence data? Ecology 83,2027–2036.

Hobbs, R.J., Huenneke, L.F., 1992. Disturbance, diversity, and invasion: implicationsfor conservation. Conservation Biology 6, 324–337.

Howden, S.M., Soussana, J.F., Tubiello, F.N., Chhetri, N., Dunlop, M., Meinke, H., 2007.Adapting agriculture to climate change. PNAS 104, 19691–19696.

Lawton, J.L., 2007. Climate change 2007: the physical science basis. In: Solomon, S.,Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L.(Eds.), Contribution of I to the Fourth Assessment Report of the Intergovern-mental Panel on Climate Change. Cambridge University Press, Cambridge,United Kingdom/New York, NY.

Lawton, J.L., 2000. Concluding remarks: a review of some open questions. In:Hutchings, et al. (Eds.), Ecological Consequences of Heterogeneity. CambridgeUniversity Press, Cambridge.

Mack, R.N., 1996. Predicting the identity and fate of plant invaders: emergent andemerging approaches. Biological Conservation 78, 107–121.

Mac Nally, R.C., 1989. The relationship between habitat breadth, habitat position,and abundance in forest and woodland birds along a continental gradient. Oikos54, 44–54.

McDonald, A.J., Riha, S.J., Mohler, C.L., 2004. Mining the record: historical evidencefor climatic influences on maize—Abutilon theophrasti competition. WeedResearch 44, 439–445.

Milchunas, D.G., Lauenroth, W.K., 1995. Inertia in plant community structure—statechanges after cessation of nutrient-enrichment stress. Ecological Applications5, 452–458.

Nakicenovic, N., Alcamo, J., Davis, G., de Vries, B., Fenhann, J., Gaffin, S., Gregory, K.,Grubler, A., Jung, T.Y., Kram, T., La Rovere, E.L., Michaelis, L., Mori, S., Morita, T.,Pepper, W., Pitcher, H., Price, L., Riahi, K., Roehrl, A., Rogner, H.H., Sankovski, A.,Schlesinger, M., Shukla, P., Smith, S., Swart, R., van Rooijen, S., Victor, N., Dadi, Z.,2000. IPCC Special Report on Emissions Scenarios. Cambridge University Press,Cambridge, UK/New York, NY.

NASS, 1992. 1992 Census of Agriculture. (http://www.agcensus.usda.gov/).Nix, H.A., 1986. A biogeogaphic analysis of Australian Elapid snakes. In: R. Longmore

(ed.), Atlas of Australian Elapid Snakes. Australian Flora and Fauna Series 8, pp.4–15.

O’Donnell, C.C., Adkins, S.W., 2001. Wild oat and climate change: the effect of CO2

concentration, temperature, and water deficit on the growth and developmentof wild oat in monoculture. Weed Science 49, 694–702.

Patterson, D.T., 1995. Weeds in a changing climate. Weed Science 43, 685–700.Patterson, D.T., Flint, E.P., 1979. Effects of chilling on cotton (Gossypium hirsutum),

velvetleaf (Abutilon theophrasti), and spurred anoda (Anoda cristata). WeedScience 27, 473–479.

Patterson, D.T., Highsmith, M.T., Flint, E.P., 1988. Effects of temperature and CO2

concentration on the growth of cotton (Gossypium hirsutum), spurred anoda(Anoda cristata), and velvetleaf (Abutilon theophrasti). Weed Science 36, 751–757.

Patterson, D.T., Westbrook, J.K., Joyce, R.J.V., Lingren, P.D., Rogasik, J., 1999. Weeds,insects, and diseases. Climate Change 43, 711–727.

Pearson, R.G., Dawson, T.P., 2003. Predicting the impacts of climate change on thedistribution of species: are bioclimatic envelope models useful? Global Ecologyand Biogeography 12, 361–371.

Pearson, R.G., Dawson, T.P., Berry, P.M., Harrison, P.A., 2002. SPECIES: a spatialevaluation of climate impact on the envelope of species. Ecological Modeling154, 289–300.

Peterson, A.T., Papes, M., Kluza, D.A., 2003. Predicting the potential invasive distribu-tions of four alien plant species in North America. Weed Science 51, 863–868.

Pope, V.D., Gallani, M.L., Rowntree, P.R., Stratton, R.A., 2000. The impact of newphysical parameterizations in the Hadley Centre climate model—HadCM3.Climate Dynamics 16, 123–146.

Quinn, G.P., Keough, M.J., 2002. Experimental Design and Data Analysis for Biol-ogists. Cambridge University Press, Cambridge.

Raimundo, R.L.G., Fonseca, R.L., Schachetti-Pereira, R., Peterson, A.T., Lewinsohn,T.M., 2007. Native and exotic distributions of siamweed (Chromolaena odorata)modeled using the Genetic Algorithm for Rule-Set Production. Weed Science 55,41–48.

Saebo, A., Mortensen, L.M., 1998. Influence of elevated atmospheric CO2 concen-tration on common weeds in Scandinavian agriculture. Acta AgriculturaeScandinavica 48, 138–143.

Salisbury, E.J., 1961. Weeds and Aliens. Collins Publishers, London.Stockwell, D.R.B., Peters, D.P., 1999. The GARP modeling system: problems and

solutions to automated spatial prediction. International Journal of GeographicInformation Systems 13, 143–158.

Stoller, E.W., 1973. Effect of minimum soil temperature on differential distributionof Cyperus rotundus and C. esculentus in the United States. Weed Research 13,209–217.

Sutherst, R.W., Maywald, G.F., 1985. A computerized system for matching climatesin ecology. Agriculture, Ecosystems & Environment 30, 805–816.

Thuiller, W., 2003. BIOMOD—optimizing predictions of species distributions andprojecting potential future shifts under global change. Global Change Biology 9,1353–1362.

Tubiello, F.N., Soussana, J.F., Howden, S.M., 2007. Crop and pasture response toclimate change. PNAS 104, 19686–19690.

Tungate, K.D., Israel, D.W., Watson, D.M., Rufty, T.W., 2007. Potential changes inweed competitiveness in an agroecological system with elevated temperature.Environmental and Experimental Botany 60, 42–49.

Washington, W.M., Weatherly, J.W., Meehl, G.A., Semtner Jr., A.J., Bettge, T.W., Craig,A.P., Strand Jr., W.G., Arblaster, J., Wayland, V.B., James, R., Zhang, Y., 2000.Parallel Climate Model (PCM) control and transient simulations. ClimateDynamics 16, 755–774.

Woodward, F.I., 1987. Climate and Plant Distribution. Cambridge University Press,Cambridge.

Woodward, F.I., 1988. Temperature and the distribution of plant species. Symposiaof the Society for Experimental Biology 42, 59–75.

Zhu, C., Zeng, Q., Ziska, L.H., Zhu, J., Xie, Z., Liu, G., 2008. Effect of nitrogen supply oncarbon dioxide-induced changes in competition between rice and barnyard-grass (Echinocholoa crus-galli). Weed Science 56, 66–71.

Ziska, L.H., 2000. The impact of elevated CO2 on yield loss from a C3 and C4 weed infield-grown soybean. Global Change Biology 6, 899–905.

Ziska, L.H., 2001. Changes in competitive ability between a C4 crop and a C3 weedwith elevated carbon dioxide. Weed Science 49, 622–627.

Ziska, L.H., 2002. Influence of rising atmospheric CO2 since 1900 on early growthand photosynthetic response of a noxious invasive weed, Canada thistle (Cir-sium arvense). Functional Plant Biology 29, 1387–1392.

Ziska, L.H., 2003. Evaluation of yield loss in field sorghum from a C3 and C4 weedwith increasing CO2. Weed Science 51, 914–918.

Ziska, L.H., Faulkner, S.S., Lydon, J., 2004. Changes in biomass and root:shoot ratio ina field-grown, noxious perennial weed, Canada thistle (Cirsium arvense (L.)Scop.) with elevated CO2: implications for chemical control by glyphosate.Weed Science 52, 584–588.

Ziska, L.H., Teasdale, J.R., 2000. Sustained growth and increased tolerance toglyphosate observed in a C3 perennial weed, quackgrass (Elytrigia repens),grown at elevated carbon dioxide. Australian Journal of Plant Physiology 27,159–166.